I really appreciate your responses. But I can't see how that would work (tested it too). That would, if I understand it correctly, ease in the Radius Change value. So it would start at zero and go to 1cm in my example. But this would result in an end Radius Change value of 1cm and lead to a very hard end of the scale animation. So there is a difference in easing the scale amount (Radius Value) and the amount in which it is scaled (Radius Change). Also it would not care about the random sizes, so smaller and larger particles would ease differently.
Again, I might be totally wrong, but it still isn't working. It is of course being slower in the beginning (ease in) but not in the end (ease out).